Progress Toward Polio Eradication — Worldwide, January 2019–June 2021

In 1988, when the Global Polio Eradication Initiative (GPEI) began, polio paralyzed >350,000 children across 125 countries. Today, only one of three wild poliovirus serotypes, type 1 (WPV1), remains in circulation in only two countries, Afghanistan and Pakistan. This report summarizes progress toward global polio eradication during January 1, 2019-June 30, 2021 and updates previous reports (1,2). In 2020, 140 cases of WPV1 were reported, including 56 in Afghanistan (a 93% increase from 29 cases in 2019) and 84 in Pakistan (a 43% decrease from 147 cases in 2019). As GPEI focuses on the last endemic WPV reservoirs, poliomyelitis outbreaks caused by circulating vaccine-derived poliovirus (cVDPV) have emerged as a result of attenuated oral poliovirus vaccine (OPV) virus regaining neurovirulence after prolonged circulation in underimmunized populations (3). In 2020, 32 countries reported cVDPV outbreaks (four type 1 [cVDPV1], 26 type 2 [cVDPV2] and two with outbreaks of both); 13 of these countries reported new outbreaks. The updated GPEI Polio Eradication Strategy 2022-2026 (4) includes expanded use of the type 2 novel oral poliovirus vaccine (nOPV2) to avoid new emergences of cVDPV2 during outbreak responses (3). The new strategy deploys other tactics, such as increased national accountability, and focused investments for overcoming the remaining barriers to eradication, including program disruptions and setbacks caused by the COVID-19 pandemic.

In 1988, when the Global Polio Eradication Initiative (GPEI) began, polio paralyzed >350,000 children across 125 countries. Today, only one of three wild poliovirus serotypes, type 1 (WPV1), remains in circulation in only two countries, Afghanistan and Pakistan. This report summarizes progress toward global polio eradication during January 1, 2019-June 30, 2021 and updates previous reports (1,2). In 2020, 140 cases of WPV1 were reported, including 56 in Afghanistan (a 93% increase from 29 cases in 2019) and 84 in Pakistan (a 43% decrease from 147 cases in 2019). As GPEI focuses on the last endemic WPV reservoirs, poliomyelitis outbreaks caused by circulating vaccine-derived poliovirus (cVDPV) have emerged as a result of attenuated oral poliovirus vaccine (OPV) virus regaining neurovirulence after prolonged circulation in underimmunized populations (3). In 2020, 32 countries reported cVDPV outbreaks (four type 1 [cVDPV1], 26 type 2 [cVDPV2] and two with outbreaks of both); 13 of these countries reported new outbreaks. The updated GPEI Polio Eradication Strategy 2022-2026 includes expanded use of the type 2 novel oral poliovirus vaccine (nOPV2) to avoid new emergences of cVDPV2 during outbreak responses (3). The new strategy deploys other tactics, such as increased national accountability, and focused investments for overcoming the remaining barriers to eradication, including program disruptions and setbacks caused by the COVID-19 pandemic.

Polio Vaccination
In worldwide immunization programs, OPV and at least 1 dose of injectable, inactivated poliovirus vaccine (IPV) are routinely used. Because IPV contains all three poliovirus serotypes, it protects against disease in children who seroconvert after vaccination; however, it does not prevent poliovirus transmission. In 2016, a global coordinated switch occurred from trivalent OPV (tOPV), which contains Sabin strain types 1, 2, and 3 to bivalent OPV (bOPV), which contains Sabin strain types 1 and 3. WPV2 was declared eradicated in 2015, and cVDPV2 was the predominant cause of cVDPV outbreaks after the last WPV2 case was detected in 1999. The use of monovalent OPV Sabin strain type 2 (mOPV2) is reserved for cVDPV2 outbreak responses. In November 2020, the World Health Organization (WHO) granted Emergency Use Listing (EUL) for genetically stabilized nOPV2 to be used in a limited number of countries that have met readiness criteria for initial use* of nOPV2 (5) in response to outbreaks.
In 2020, the estimated global infant coverage with 3 doses of poliovirus vaccine (Pol3) by age 1 year was 83% (6). However, substantial variation in coverage exists by WHO region, nationally, and subnationally. In the two countries with endemic WPV (Afghanistan and Pakistan), 2020 POL3 coverage was 75% and 83%, respectively (6); estimated coverage in subnational areas with transmission is much lower.
In 2019, GPEI supported 199 supplementary immunization activities (SIAs) † in 42 countries with approximately 1 billion bOPV, 20 million IPV, 32 million monovalent OPV type 1 (mOPV1), and 142 million mOPV2 doses administered. In 2020, 149 SIAs were conducted in 30 countries with approximately 696 million bOPV, 6 million IPV, 4 million mOPV1, * Authorization for wider use under EUL is pending review of safety and effectiveness data from the initial use. † Mass immunization campaigns intended to interrupt poliovirus circulation by immunizing every child aged <5 years with 2 OPV doses, regardless of previous immunization status. 228 million mOPV2, and 51 million tOPV doses administered; tOPV was used during four SIAs in Afghanistan and Pakistan, where cocirculation of WPV1 and cVDPV2 requires tOPV for efficiency in scheduling and implementing SIAs; GPEI authorized restarting filling of tOPV stocks for this purpose. In 2021, approximately 136 million nOPV2 doses have been released in eight countries approved for initial use (Benin, Chad, Congo, Liberia, Niger, Nigeria, Sierra Leone, and Tajikistan). SIAs continue to be affected by the COVID-19 pandemic § in 2021.

Poliovirus Surveillance
WPV and cVDPV transmission are detected primarily through surveillance for acute flaccid paralysis (AFP) among children aged <15 years with testing of stool specimens at one of 145 WHO-accredited laboratories of the Global Polio Laboratory Network (7). During January-September 2020, the number of reported AFP cases declined 33% compared with the same period in 2019 (8). Environmental surveillance (testing of sewage for poliovirus) can supplement AFP surveillance; however, environmental sampling also declined somewhat during this period. Current data indicate that the COVID-19 pandemic has continued to limit AFP surveillance sensitivity.
were found across 38 districts compared with 20 districts in 2019. As of August 3, 2021, one WPV1 case was reported in Afghanistan in 2021, a 97% decrease compared with the first 6 months of 2020. Pakistan reported 84 WPV1 cases from 39 districts in 2020, representing a 43% decrease from the 147 cases reported in 43 districts during 2019. One WPV1 case has been reported during January-June 2021, from Balochistan province, a 98% decrease from the 60 WPV1 cases from five provinces during the same 2020 period. This period accounted for 71% of all Pakistan WPV1 cases in 2020. In both countries, the number of orphan WPV1 isolates (those with ≤98.5% genetic identity with other isolates) from AFP cases increased from five of 176 (3%) in 2019 to 18 of 140 (13%) in 2020, signifying an increase in AFP surveillance gaps in 2020 (7).
Environmental surveillance in Afghanistan detected WPV1 in 35 (8%) of 418 sewage samples collected during 2020 and in 57 (22%) of 264 samples in 2019 (Table 2). In Pakistan, WPV1 was detected in 434 (52%) of 830 sewage samples  (2 Countries reporting cVDPV cases and isolations. During January 2019-June 2021, cVDPV transmission was identified in 32 countries; 13 countries were affected by new cVDPV outbreaks in 2020. Afghanistan reported 308 cVDPV2 cases in 2020 compared with no cases in 2019. Pakistan reported 135 cVDPV2 cases in 2020, more than a fivefold increase from the 22 reported in 2019. To date in 2021,195 cVDPV2 cases have been identified globally, including 43 in Afghanistan and eight in Pakistan.

Discussion
With the August 2020 certification of the African Region as WPV-free,** five of the six WHO regions, representing over 90% of the world's population, are now free of wild polioviruses. Given this achievement, GPEI is focusing efforts on two goals: interrupting persistent WPV1 transmission in Pakistan and Afghanistan and stopping all current outbreaks of cVDPV2. To reach these goals, in June 2021, GPEI released a revised 5-year strategy for polio eradication that aims to address persistent challenges and recover from setbacks exacerbated by the COVID-19 pandemic (4).
In Afghanistan, the main challenges to ending poliovirus transmission are the inability to reach all children in critical areas near reservoirs in Pakistan and increasing political instability. The polio program in Afghanistan has continued to operate for many years, even during periods of insecurity and escalating conflict. Although negotiations with local leaders in Afghanistan facilitated vaccination efforts at one time, restrictions on vaccinations have persisted in areas controlled by insurgent groups since the October 2018 ban on house-to-house campaigns, which has

Summary
What is already known about this topic?
What is added by this report?
From 2019 to 2020, the number of WPV1 cases increased in Afghanistan and decreased in Pakistan and the number of cVDPV2 cases increased and cVDPV2 outbreak countries increased to 32. In Afghanistan, the polio program faces challenges including an inability to reach children in critical areas and increasing political instability. The COVID-19 pandemic continues to limit the quality of immunization activities and poliovirus surveillance.
What are the implications for public health practice?
The Polio Eradication Strategy for 2022-2026 outlines measures including increased government accountability and wider use of novel, oral poliovirus vaccine type 2 that are needed to eradicate polio. since expanded geographically (10). WHO is anticipating that some negotiated access will again be possible. Other challenges include current mass population movements, clusters of vaccine refusals, and suboptimal SIA quality in some areas previously under government control (10).
Globally, cVDPV2 outbreaks increased in number and geographic extent during 2019-2020 because of delays in mOPV2 response SIAs, which were frequently of low quality. Since the switch in 2016 from tOPV to bOPV, 1,755 cases of paralytic polio have been reported from 64 cVDPV2 outbreaks in 30 countries across four WHO regions (4). † † GPEI has outlined a strategy for stopping cVDPV transmission and reducing the risk of seeding new outbreaks by expanding use of nOPV2 (4). Continued monitoring will be needed to ensure safety and effectiveness while nOPV2 is brought into wider use and to ascertain whether it can replace mOPV2 (5).
The findings in this report are subject to at least one limitation. SIAs, field surveillance, and investigation activities were curtailed in 2020 because of COVID-19 pandemic mitigation measures, and laboratory testing suffered delays (8); limitations on SIA quality and surveillance sensitivity continue in 2021. On the other hand, the COVID-19 pandemic has presented opportunities to jointly increase the effectiveness of polio eradication activities and promote health services integration. For example, the global rollout of COVID-19 vaccines presents an opportunity to strengthen demand for vaccination against both COVID-19 and polio. † † https://polioeradication.org/polio-today/polio-now/this-week/ circulating-vaccine-derived-poliovirus/ Thousands of polio eradication workers worldwide continue to play a critical role in implementing countries' COVID-19 responses. Maintaining these partnerships will be important in eradicating WPV and stopping cVDPV transmission while simultaneously addressing other health priorities. 8 Nonfatal and fatal drug overdoses increased overall from 2019 to 2020 (1).* Illicit benzodiazepines (e.g., etizolam, flualprazolam, and flubromazolam) † were increasingly detected among postmortem and clinical samples in 2020, often with opioids, § and might have contributed to overall increases in drug overdoses. Availability of recent multistate trend data on nonfatal benzodiazepine-involved overdoses and involvement of illicit benzodiazepines in overdoses is limited. This data gap was addressed by analyzing annual and quarterly trends in suspected benzodiazepine-involved nonfatal overdoses ¶ treated in emergency departments (EDs) (benzodiazepine overdose ED visits) during January 2019-December 2020 (32 states and the District of Columbia [DC]) and benzodiazepineinvolved overdose deaths (benzodiazepine deaths), which include both illicit and prescription benzodiazepines, during January 2019-June 2020 (23 states) from CDC's Overdose Data to Action (OD2A) program. From 2019 to 2020, benzodiazepine overdose ED visits per 100,000 ED visits increased (23.7%), both with opioid involvement (34.4%) and without (21.0%). From April-June 2019 to April-June 2020, overall benzodiazepine deaths increased 42.9% (from 1,004 to 1,435), prescription benzodiazepine deaths increased 21.8% (from 921 to 1,122), and illicit benzodiazepine deaths increased 519.6% (from 51 to 316). During January-June 2020, most (92.7%) benzodiazepine deaths also involved opioids, mainly illicitly manufactured fentanyls (IMFs) (66.7%). Improving naloxone availability and enhancing treatment access for persons using benzodiazepines and opioids and calling emergency services for overdoses involving benzodiazepines and opioids, coupled with primary prevention of drug use and misuse, could reduce morbidity and mortality. * https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm † "Illicit benzodiazepines" refers to benzodiazepines that are not marketed in the United States for medical purposes; "prescription benzodiazepines" refers to those that are marketed as prescription drugs in the United States (but does not imply that the decedent had a prescription for that benzodiazepine; prescription benzodiazepines might be diverted). https://www.who.int/ medicines/access/controlled-substances/Final_Etizolam.pdf; https://www. npsdiscovery.org/wp-content/uploads/2019/12/Public-Alert_Flualprazolam_ NPS-Discovery_120519.pdf; https://www.who.int/docs/default-source/ controlled-substances/43rd-ecdd/final-flubromazolam-a.pdf?sfvrsn § https://www.npsdiscovery.org/reports/trend-reports/ ¶ Analyses were intended to include nonfatal overdose visits with unintentional and undetermined intents. ED visits resulting in death were not excluded but accounted for < 0.5% of total benzodiazepine overdose ED visits in ESSENCE during the study period.
CDC's OD2A program collects data on unintentional and undetermined intent drug overdoses: 1) nonfatal overdoses treated in EDs from the Drug Overdose Surveillance and Epidemiology (DOSE) system, and 2) overdose deaths from the State Unintentional Drug Overdose Reporting System (SUDORS).** Benzodiazepine overdose ED visits during January 2019-December 2020 were identified from 33 DOSE jurisdictions (32 states and DC) † † submitting data to the National Syndromic Surveillance Program. Benzodiazepine and opioid overdose ED visits were identified using diagnosis codes and chief complaint text fields. § § Only EDs consistently reporting informative data ¶ ¶ during January 2019-December 2020 were included to ensure valid trend analyses. Relative rate percentage changes for benzodiazepine overdose ED visits per 100,000 ED visits were calculated by quarter (Q1: January-March, Q2: April-June, Q3: July-September, and Q4: October-December) and stratified by opioid involvement.*** Benzodiazepine deaths were identified from 23 states † † † participating in SUDORS. States obtained data from death certificates and medical examiner and coroner reports, including complete postmortem toxicology testing results. Benzodiazepine deaths during January 2019-June 2020, percentages involving any opioids and opioid type (heroin, IMFs, or prescription opioids), § § § and percentage change in deaths, were calculated by quarter. Demographic characteristics of persons experiencing nonfatal and fatal benzodiazepine overdoses, and specific drug co-involvement for benzodiazepine deaths, were described using the most recent period of available data (benzodiazepine overdose ED visits: January-December 2020, benzodiazepine deaths: January-June 2020). Benzodiazepine deaths were § § § Drugs coded as prescription opioids were alfentanil, buprenorphine, codeine, hydrocodone, hydromorphone, levorphanol, loperamide, meperidine, methadone, morphine, noscapine, oxycodone, oxymorphone, pentazocine, prescription fentanyl, propoxyphene, remifentanil, sufentanil, tapentadol, and tramadol. IMFs include illicitly manufactured fentanyl and illicit fentanyl analogs. Fentanyl was classified as likely illicitly manufactured or likely prescription using toxicology, scene, and witness evidence. In the absence of sufficient evidence to classify fentanyl as illicit or prescription (<7% of deaths involving fentanyl), it was classified as illicit because most fentanyl overdose deaths involve illicit fentanyl.
stratified by benzodiazepine type (prescription or illicit). Chi-square tests were used for pairwise comparisons; p-values <0.05 were considered statistically significant. Analyses were conducted using SAS (version 9.4; SAS Institute). This activity was reviewed by CDC and conducted consistent with applicable federal law and CDC policy. ¶ ¶ ¶ During January 2019-December 2020, 117 million ED visits reported in the 33 jurisdictions (72% of total ED visits) qualified for inclusion in analyses. Among these, 31,377 benzodiazepine overdose ED visits were identified, including 15,547 in 2019 and 15,830 in 2020; 6,883 (21.9%) also involved opioids. The highest number of benzodiazepine overdose ED visits occurred in Q3 2020 (4,181) ( Figure 1).
In 2020, benzodiazepine overdose ED visits more often involved females (51.5%) and persons aged 25-34 years   Abbreviations: DOSE = Drug Overdose Surveillance and Epidemiology; IMFs = illicitly manufactured fentanyls; IQR = interquartile range; N/A = data not available; OD = overdose; Q = quarter; SUDORS = State Unintentional Drug Overdose Reporting System. * Nonfatal benzodiazepine overdose data were from 32 states and the District of Columbia for overdose emergency department visits during January 1, 2020-December 31, 2020. Fatal benzodiazepine overdose data were from 23 states for deaths during January 1, 2020-June 30, 2020. † Numbers of any benzodiazepines fatal ODs will not reflect the sum of prescription plus illicit benzodiazepine fatal ODs because some deaths involved both, and some deaths had a generic listing of benzodiazepine involvement; therefore, the prescription or illicit status could not be classified. § Dashes indicate cell data suppressed because cell contains one to nine cases or to prevent calculation of other suppressed cells. ¶ Comparing prescription benzodiazepine overdose deaths to illicit benzodiazepine overdose deaths, the percent co-involvement of any opioids was not statistically significantly different at p<0.05; all other comparisons were statistically significantly different (sex, age, race/ethnicity, co-involvement of heroin, IMFs, illicit opioids, prescriptions opioids, prescription and illicit opioids, and benzodiazepine type). Because prescription/illicit benzodiazepine deaths were not mutually exclusive, chi-square testing was performed after excluding 79 deaths with both prescription and illicit benzodiazepine involvement; however, percentages of each demographic category and with each substance co-involvement were similar to the nonmutually exclusive categorizations. ** IMFs include fentanyl and illicit fentanyl analogs. † † Illicit opioids include heroin, IMFs, and other non-fentanyl illicit synthetic opioids (e.g., isotonitazene). § § Prescription opioids include alfentanil, buprenorphine, codeine, hydrocodone, hydromorphone, levorphanol, loperamide, meperidine, methadone, morphine, noscapine, oxycodone, oxymorphone, pentazocine, prescription fentanyl, propoxyphene, remifentanil, sufentanil, tapentadol, and tramadol.

Discussion
This is the first multistate report to examine recent trends in both nonfatal and fatal benzodiazepine overdoses. Three concerning trends during 2019-2020 were identified: 1) increases in both nonfatal and fatal overdoses involving benzodiazepines and opioids; 2) marked increases in illicit benzodiazepine deaths, although overdose deaths involving prescription benzodiazepines still far outnumber those involving illicit benzodiazepines; and 3) increases in nonfatal benzodiazepine overdoses not involving opioids. In 2016, the CDC Opioid Prescribing Guideline discouraged co-prescribing opioids and benzodiazepines,**** and % co-heroin % co-opioid % co-IMFs % co-Rx opioid

Illicit benzodiazepine
Rx benzodiazepine Any benzodiazepine the Food and Drug Administration imposed its most prominent warning on all benzodiazepine medications, † † † † describing the risks of use with opioids. Despite progress in reducing coprescribing before 2019 (2), this study suggests a reversal in the decline in benzodiazepine deaths from 2017 to 2019, § § § § driven in part by increasing involvement of IMFs in benzodiazepine deaths and influxes of illicit benzodiazepines, likely indicating simultaneous use of nonprescribed opioids and benzodiazepines. During 2019-2020, benzodiazepine deaths (both prescription and illicit) were characterized by high and increasing coinvolvement of IMFs, a trend documented as early as during 2017-2018 (3). Substantial increases in the supply of IMFs during January 2013-June 2020, ¶ ¶ ¶ ¶ coupled with the high potency and rapid absorption of IMFs (4), which increase overdose risk above that of heroin and prescription opioids, is likely a principal driver of fatal benzodiazepine and IMF overdose. The largest increase in IMF involvement among benzodiazepine deaths occurred in 2020 between Q1 and Q2, possibly reflecting altered drug use patterns that increased overdose risk (e.g., decreased naloxone access); and possible drug supply disruptions; during the COVID-19 pandemic (5). Although the much greater involvement of opioids in benzodiazepine deaths (91.4%) compared with benzodiazepine overdose ED visits (21.9%) underscores the dangers of co-use, increases in opioid involvement among benzodiazepine ED visits (34.4% increase) throughout 2020 might be an early indicator of continued and amplified increases in morbidity and mortality related to benzodiazepine and opioid co-use.
Other factors accelerating increases in benzodiazepine deaths involving opioids are rapid increases in supply and co-use of illicit benzodiazepines among persons using illicit opioids, especially IMFs. Whereas law enforcement reports of diverted prescription benzodiazepines declined from 2015 through June 2020, reports of illicit benzodiazepines (particularly etizolam and flualprazolam) surged during that period, indicating increased availability (6).***** Reductions in benzodiazepine and opioid co-prescribing must be coupled with efforts to disrupt and reduce the availability of and harms associated with concurrent use of illicit benzodiazepines and IMFs.
Although rates of ED visits for mental health conditions increased during 2019-2020 (1), benzodiazepine prescriptions † † † † https://www.fda.gov/drugs/drug-safety-and-availability/fda-drugsafety-communication-fda-warns-about-serious-risks-and-deathwhen-combining-opioid-pain-or; https://www.fda.gov/drugs/ drug-safety-and-availability/fda-requiring-boxed-warning-updatedimprove-safe-use-benzodiazepine-drug-class § § § § https://www.drugabuse.gov/drug-topics/trends-statistics/overdose-deathrates; https://wonder.cdc.gov/ ¶ ¶ ¶ ¶ h t t p s : / / w w w. n f l i s . d e a d i v e r s i o n . u s d o j . g o v / n f l i s d a t a / docs/13915NFLISdrugMidYear2020.pdf ***** h t t p s : / / w w w. n f l i s . d e a d i v e r s i o n . u s d o j . g o v / n f l i s d a t a / docs/13915NFLISdrugMidYear2020.pdf remained relatively stable during January 2019-May 2020, with a transient spike in March 2020 indicating, per recommendations, increases in the availability of medications on hand because of stay-at-home orders to slow the spread of COVID-19 (7). However, the increases in benzodiazepine overdose ED visit rates, including those without opioids, raise concerns about increased misuse and warrant further investigation. Because benzodiazepine use is less likely to result in fatal overdose without use of opioids or other depressants (6), tracking nonfatal benzodiazepine overdoses is critical to tracking benzodiazepine misuse trends. The findings in this report are subject to at least five limitations. First, jurisdictions included in nonfatal and fatal overdose analyses are not nationally representative and differ from each other, limiting the extent to which trends can be compared. Second, full toxicology results for nonfatal overdoses were not available, and opioid and benzodiazepine involvement in nonfatal overdoses is likely underestimated because comprehensive toxicology testing of persons treated for overdoses varies within and across EDs, and hospital discharge codes with drug specific information might be unavailable (8). Third, despite only including consistently reporting facilities, ED visits decreased sharply after implementation of COVID-19 mitigation measures in March 2020, which might inflate rate increases (9). Fourth, four states (Illinois, Missouri, Pennsylvania, and Wisconsin) reported overdose deaths from varying subsets of counties. Results were similar with and without these states. Finally, postmortem toxicology testing and drug involvement determination vary over time and across states, potentially affecting detection of specific drugs involved in deaths.
Increases in benzodiazepine overdose ED visits throughout 2020, coupled with increases in illicit benzodiazepine deaths since 2019, highlight the need to enhance efforts to mitigate harm from simultaneously using benzodiazepines and opioids and monitor the magnitude and persistence of increases in illicit benzodiazepine deaths. Persons who co-use opioids and benzodiazepines might be less likely to receive medications for opioid use disorder than persons using opioids only (10); therefore, efforts to increase treatment access should be enhanced. Expansion of naloxone availability and rapid naloxone administration should be encouraged for overdoses involving benzodiazepines and opioids because naloxone reverses opioid overdoses irrespective of benzodiazepine presence. However, educational efforts should emphasize the dangers of using illicit benzodiazepines, especially in combination with opioids, and the importance of calling 9-1-1 even after naloxone administration, because benzodiazepine overdose symptoms are unaffected by naloxone and might require additional medical treatment. These efforts, complemented by broader primary prevention of drug use and misuse, could prevent drug overdose morbidity and mortality.

Summary
What is already known about this topic?
What is added by this report?
What are the implications for public health practice?
Improving naloxone availability and enhancing treatment access for persons using benzodiazepines and opioids and calling emergency services for overdoses involving benzodiazepines and opioids, coupled with primary prevention of drug use and misuse, could reduce morbidity and mortality.

Mental Health and Substance Use Among Adults with Disabilities During the COVID-19 Pandemic -United States, February-March 2021
Mark É. Czeisler 1,2,3,4  Adults with disabilities, a group including >25% of U.S. adults (1), experience higher levels of mental health and substance use conditions and lower treatment rates than do adults without disabilities* (2,3). Survey data collected during April-September 2020 revealed elevated adverse mental health symptoms among adults with disabilities (4) compared with the general adult population (5). Despite disproportionate risk for infection with SARS-CoV-2, the virus that causes COVID-19, and COVID-19-associated hospitalization and mortality among some adults with disabilities (6), information about mental health and substance use in this population during the pandemic is limited. To identify factors associated with adverse mental health symptoms and substance use among adults with disabilities, the COVID-19 Outbreak Public Evaluation (COPE) Initiative † administered nonprobability-based Internet surveys to 5,256 U.S. adults during February-March 2021 (response rate = 62.1%). Among 5,119 respondents who completed a two-item disability screener, nearly one-third (1,648; 32.2%) screened as adults with disabilities. These adults more frequently experienced symptoms of anxiety or depression (56.6% versus 28.7%, respectively), new or increased substance use (38.8% versus 17.5%), and suicidal ideation (30.6% versus 8.3%) than did adults without disabilities. Among all adults who had received a diagnosis of mental health or substance use conditions, adults with disabilities more frequently (42.6% versus 35.3%; p <0.001) reported that the pandemic made it harder for them to access related care or medication. Enhanced mental health and substance use screening among adults with disabilities and improved access to medical services are critical during public health emergencies such as the COVID-19 pandemic.
During February 16-March 8, 2021, among 8,475 eligible invited respondents aged ≥18 years, 5,261 (62.1%) completed nonprobability based, English-language, Internet-based * https://store.samhsa.gov/sites/default/files/d7/priv/pep19-02-00-002_508_022620.pdf † The COVID-19 Outbreak Public Evaluation (COPE) Initiative (https://www. thecopeinitiative.org/) is designed to assess public attitudes, behaviors, and beliefs related to COVID-19 pandemic and to evaluate mental and behavioral health during the pandemic. The COPE Initiative surveys included in this analysis were administered by Qualtrics, LLC (https://www.qualtrics.com), a commercial survey company with a network of participant pools with varying recruitment methodologies that include digital advertisements and promotions, word-of-mouth and membership referrals, social networks, television and radio advertisements, and offline mail-based approaches.
Qualtrics surveys for COPE. § Participants provided informed consent electronically. Quota sampling and survey weighting were used to match U.S. Census Bureau's 2019 American Community Survey adult U.S. population estimates for sex, age, and race/ethnicity to enhance the representativeness of this nonrandom sample.
Among 5,256 respondents who answered questions for weighting variables, 5,119 (97.4%) completed a two-question disability screener. ¶ Respondents completed clinically validated self-screening instruments for symptoms of anxiety and depression** and reported past-month new or increased substance use to cope with stress or emotions and serious suicidal ideation. † † Respondents also indicated prepandemic and past-month use of seven classes § § of substances to cope with stress or emotions. Adults with diagnosed anxiety, depression, posttraumatic stress disorder, or substance use disorders indicated whether their ability to access care or medications for these conditions was easier, harder, or unaffected because of the pandemic. Prevalence estimates for adverse mental health symptoms and substance use were compared among adults with and without disabilities using chi-square tests. Multivariable Poisson regression models with robust standard error estimators were used to estimate adjusted prevalence ratios (aPRs) by symptom type among adults with and without disabilities. To calculate associations between disability status and adverse § Eligibility to complete surveys was determined after electronic contact of potential participants with inclusion criteria of age ≥18 years and residence within the United States. ¶ Disability was defined as such based on a qualifying response by an adult to either one of two questions: "Are you limited in any way in any activities because of physical, mental, or emotional condition?" and "Do you have any health conditions that require you to use special equipment, such as a cane, wheelchair, special bed, or special telephone?" https://www.cdc.gov/brfss/ questionnaires/pdf-ques/2015-brfss-questionnaire-12-29-14.pdf ** Symptoms of anxiety and depression were assessed with the four-item Patient Health Questionnaire (PHQ-4). Respondents who scored ≥3 out of 6 on the Generalized Anxiety Disorder (GAD-2) and Patient Health Questionnaire (PHQ-2) subscales were considered symptomatic for the respective conditions. † † New or increased substance use was assessed with the question, "Have you started or increased using substances to help you cope with stress or emotions during the COVID-19 pandemic? Substance use includes alcohol, legal or illegal drugs, or prescription drug use in any way not directed by a doctor." Suicidal ideation was assessed with an item from the National Survey on Drug Use and Health (https://nsduhweb.rti.org/respweb/homuepage.cfm) adapted to refer to the preceding 30 days, "At any time in the past 30 days, did you seriously think about trying to kill yourself?" § § Alcohol, marijuana, cocaine, methamphetamine, prescription or illicit opioids, benzodiazepines, and prescription drugs other than opioids used in a way not directed by a doctor.  (Table). Overall, 64.1% of adults with disabilities reported adverse mental health symptoms or substance use compared with 36.0% of adults without disabilities; past-month substance use was higher among adults with disabilities (40.6%) than among adults without disabilities (24.5%). Prevalence estimates of each of the following were higher among adults with disabilities than among adults without disabilities: symptoms of anxiety or depression (56.6% versus 28.7%, respectively), new or increased substance use (38.8% versus 17.5%), and serious suicidal ideation (30.6% versus 8.3%) (Supplementary Table, https://stacks.cdc.gov/view/cdc/108999). At all timepoints, aPRs for all symptom types were significantly higher among adults with disabilities than among adults without disabilities ( Figure 1). During February 16-March 8, 2021, among adults with disabilities, aPRs for symptoms of anxiety or depression and new or increased substance use were approximately ¶ ¶ Models to estimate aPRs for adverse mental health symptoms and substance use were run with each of the collinear variables income and education during preliminary analysis. Estimated aPRs did not differ meaningfully. In the report, the models including income were included to account for potential differences in access to health care more directly. To avoid collinearity with age, employment status was included in a separate model, and aPRs were not estimated for retired status or student employment status. *** Adults who were in parental or unpaid caregiving roles were self-identified.
For this analysis, the definition of unpaid caregivers of adults was having provided unpaid care to a relative or friend aged ≥18 years to help them take care of themselves at any time during the three months before the survey. The definition of someone in a parental role was having provided unpaid care to a relative or friend aged <18 years. Respondents were categorized as being in a parental role only, a caregiver of adults role only, having both parental and caregiving roles, or having neither parental nor caregiving roles. Adults in parenting roles might not have been biologic or adoptive parents of the children.  The prevalence of substance use to cope with stress or emotions among adults with disabilities was higher than that among adults without disabilities, both prepandemic (39.7% versus 25.3%, respectively) and in the past month (40.6% versus 24.5%; both p<0.001) ( Figure 2). Among adults with disabilities, the past-month prevalence of methamphetamine use (8.4%), nonopioid prescription drug misuse (4.9%), and polysubstance use (16.9%) was approximately twice as high, and the prevalence of cocaine use (6.4%) and prescription or illicit opioid use (9.1%) were nearly three times as high compared with those among adults without disabilities (methamphetamine use 3.4%; nonopioid prescription drug misuse 2.0%; polysubstance use 7.9%; cocaine use 2.2%; prescription or illicit opioid use 3.2%). Past-month methamphetamine use prevalence increased significantly compared with prepandemic use prevalence among all respondents (with disabilities, 45.6% increase, p<0.001; without disabilities, 40.6% increase, p = 0.003). Among respondents who reported a diagnosed mental health or substance use condition, a higher percentage of adults with (versus without) disabilities reported that accessing care or medication was harder because of the COVID-19 pandemic (42.6% versus 35.3%, respectively, p<0.001).

Discussion
Nearly two thirds of surveyed adults with disabilities (who represented approximately 32% of the sample) reported adverse mental health symptoms or substance use in early 2021, compared with approximately one third of adults without disabilities. Serious suicidal ideation was approximately 2.5 times as high among adults with disabilities, and methamphetamine use, opioid use, nonopioid prescription drug misuse, and polysubstance use were at least twice as

TABLE. (Continued) Prevalence of symptoms of anxiety or depression, substance use, and suicidal ideation among adults with disabilities, by disability status and other characteristics -United States, February 16-March 8, 2021
Abbreviations: N/A = not applicable; USD = U.S. dollars. * Weighted rounded counts and percentages might not sum to expected values. † Symptoms of anxiety and depression were assessed via the four-item Patient Health Questionnaire (PHQ-4). Respondents who scored ≥3 out of 6 on the Generalized Anxiety Disorder (GAD-2) and Patient Health Questionnaire (PHQ-2) subscales were considered symptomatic for these respective conditions. § New or increased substance use was assessed by using the question, "Have you started or increased using substances to help you cope with stress or emotions during the COVID-19 pandemic? Substance use includes alcohol, legal or illegal drugs, or prescription drug use in any way not directed by a doctor. " ¶ Suicidal ideation was assessed by using an item from the National Survey on Drug Use and Health (https://nsduhweb.rti.org/respweb/homepage.cfm) adapted to refer to the previous 30 days, "At any time in the past 30 days, did you seriously think about trying to kill yourself?" ** Adults who had a disability were defined as such based on a qualifying response to either one of two questions: "Are you limited in any way in any activities because of physical, mental, or emotional condition?" and "Do you have any health conditions that require you to use special equipment, such as a cane, wheelchair, special bed, or special telephone?" Respondents who completed only one of the two disability screening questions (limited by a physical, mental, or emotional condition: 17); limited by a health condition that requires special equipment: 12) were classified as living with only that disability. https://www.cdc.gov/brfss/questionnaires/ pdf-ques/2015-brfss-questionnaire-12-29-14.pdf † † Gender responses of "Transgender" (22; 0.4%) and "None of these" (15; 0.3%) are not shown because of small counts. § § The non-Hispanic, multiple/other race or multiple races category includes respondents who identified as not Hispanic and as more than one race or as American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander, or any other race. ¶ ¶ Household income responses of "Prefer not to say" (225) are not shown because of an inability to sufficiently characterize these responses. *** Adults who were in parental or unpaid caregiving roles were self-identified. For this analysis, the definition of unpaid caregivers of adults was having provided unpaid care to a relative or friend ≥18 years to help them take care of themselves at any time during the 3 months before the survey. The definition of someone in a parental role was having provided unpaid care to a relative or friend <18 years. Respondents answered these questions separately. During analysis, all respondents were categorized as being in a parental role only, caregivers of adults only, having both parental and caregiving roles, or having neither parental nor caregiving roles. Adults in parenting roles might not have been natural or legal parents of children in their care. † † † https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf § § § Invalid postcodes were provided by 28 respondents, for whom urbanicity was not categorized. https://www.hrsa.gov/rural-health/about-us/definition/datafiles.html prevalent among adults with disabilities. These findings suggest value in enhanced mental health screening among adults with disabilities and in ensuring accessibility of routine and crisis services, particularly given that many adults reported that the COVID-19 pandemic had reduced mental health and substance use care or medication accessibility. Mental health disparities among adults with disabilities were observed across demographic groups, highlighting the importance of ensuring access to disaster distress § § § and suicide prevention ¶ ¶ ¶ resources in this population. Important strategies to prevent persons from becoming suicidal include strengthening economic supports, promoting connectedness, and teaching coping skills.**** Health care providers could incorporate trauma-informed care, because adults with disabilities might have encountered stigma and trauma in previous health care interactions. Adults with disabilities more frequently reported prepandemic and past-month substance use to cope with stress or emotions compared with adults without disabilities. The substance with the largest increase in use was methamphetamine, which is particularly concerning given the increase in amphetamine overdoses † † † † (7). Drug overdose deaths rose in 2020, driven by synthetic opioids. § § § § Consistent with previous research, § § § Substance Abuse and Mental Health Services Administration National Helpline (https://www.samhsa.gov/find-help/national-helpline); Disaster Distress Helpline (https://www.samhsa.gov/disaster-preparedness). ¶ ¶ ¶ National Suicide Prevention Lifeline: 1-800-273-TALK for English, 1-888-628-9454 for Spanish, or Lifeline Crisis Chat (https:// suicidepreventionlifeline.org/chat/). **** https://www.cdc.gov/violenceprevention/pdf/suicideTechnicalPackage.pdf † † † † https://emergency.cdc.gov/han/2020/han00438.asp § § § § https://www.cdc.gov/media/releases/2020/p1218-overdose-deaths-covid-19.html adults with disabilities disproportionately reported opioid use and nonopioid prescription drug misuse (8), highlighting the importance of educating patients and ensuring clinician access to prescription drug monitoring programs. ¶ ¶ ¶ ¶ Nearly one in ten adults with disabilities reported past-month opioid use, and opioid use among adults without disabilities increased. Policies that reduce barriers to evidence-based treatment, including recently updated buprenorphine practice guidelines,***** might improve access. The findings in this report are subject to at least four limitations. First, self-reported mental health and substance use might be subject to social desirability biases and stigma, which could lead to underreporting. Second, because the surveys were English-language only and data were obtained using nonprobability-based sampling, despite quota sampling and survey weighting, the findings from this nonrandom sample might not be generalizable. However, the proportion and demographics of surveyed adults with disabilities were similar to those of recent samples from other sources with the same or similar screening questions (1,2,4), and prevalence estimates of symptoms of anxiety and depression were largely consistent with those from other sources for the U.S. adult population (9) and adults with disabilities (4) including the U.S. Census Bureau's probability-based Household Pulse Survey (64.3% among adults with disabilities compared with 27.4% among ¶ ¶ ¶ ¶ https://www.cdc.gov/drugoverdose/pdmp/states.html ***** https://www.federalregister.gov/documents/2021/04/28/2021-08961/ practice-guidelines-for-the-administration-of-buprenorphine-for-treatingopioid-use-disorder

FIGURE 1. (Continued) Adjusted prevalence ratios* and 95% confidence intervals † for ≥1 symptoms of adverse mental health or substance use (A), symptoms of anxiety or depression (B), new or increased substance use (C), and suicidal ideation (D) among adults with disabilities, compared with adults without disabilities (referent group) § -United States, February 16-March 8, 2021 ¶
Abbreviations: aPR = adjusted prevalence ratio; CI = confidence interval. * With 95% CIs indicated by error bars. Multivariable Poisson regression models included sex, age group in years, race/ethnicity, income, U.S. Census region, urbanicity, and parental or unpaid caregiving roles (parental roles were not assessed in June 2020; only unpaid caregiving roles were considered for this variable in the June 2020 models). Separate, additional models were run to estimate aPRs for the following employment statuses: essential worker, nonessential worker, and unemployed. Estimates were not made for retired or student employment statuses because of collinearity between these employment statuses and age. † For panels A, B, and C, the y-axis range for aPR estimates is 0-5, which contains all aPRs and 95% CIs for these panels with maximal view of differences in model estimates. For panel D, given the relative rarity of suicidal ideation among some demographic subgroups that results in wide CIs for aPR estimates, the y-axis range is 0-10. § Within each subgroup, adults without disabilities are the reference group used to estimate aPRs for outcomes among adults with disabilities. ¶ Estimated aPRs are during February 16-March 8, 2021, except for the "over time" estimates, which also include estimates based on data collected during June 24-30, 2020, August 28- September 6, 2020, andDecember 6-27, 2020. adults without disabilities in April 2021). † † † † † Third, the respondents with disabilities might not be representative of all adults with disabilities, some of whom might lack access to hardware or assistive technologies required to independently complete the survey. Finally, adverse mental health symptoms might, in some cases, represent respondents' disabling mental health conditions, which could confound associations with other comorbid disabling conditions (e.g., physical, cognitive, sensory); however, sensitivity analyses excluding adults with disabilities who had mental health or substance use diagnoses yielded consistent findings.
Adults with disabilities have been disproportionately affected by adverse mental health symptoms and substance use during the COVID-19 pandemic, highlighting the importance of improved access to treatment for this population. Clinicians might consider screening all patients for mental health and substance use conditions during and after the pandemic. § § § § § Behavioral health care providers might also consider facility, policy, and procedural pathway analyses to ensure accessibility for clients with physical, sensory, or cognitive disabilities. ¶ ¶ ¶ ¶ ¶ Strategies designed to increase access to care and medication during public health emergencies, such as telehealth, might consider telemedicine platform and system accessibility for adults with disabilities (10); further research to identify and address health disparities among adults with disabilities could help guide additional evidence-based strategies. † † † † † https://www.cdc.gov/nchs/covid19/pulse/functioning-and-disability.htm § § § § § https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/ drug-use-illicit-screening; https://www.uspreventiveservicestaskforce.org/ uspstf/recommendation/depression-in-adults-screening ¶ ¶ ¶ ¶ ¶ http://cct.org/wp-content/uploads/2015/08/2015ADAComplianceGuide.pdf Adults with disabilities, past-month use Adults with disabilities, prepandemic use Adults without disabilities, past-month use Adults without disabilities, prepandemic use * Overall, prepandemic and past-month use of any of these substances were reported by 39.7% and 40.6%, respectively, of adults with disabilities, and by 25.3% and 24.5%, respectively, of adults without disabilities. † All differences between adults with disabilities and adults without disabilities were significant (chi-square p-value <0.05). § Circles for use of marijuana (among adults with disabilities), use of prescription drugs (among adults without disabilities), and polysubstance use (among adults with disabilities) might appear overlapping because of very small changes in reported prevalence (<1% in all cases).

Summary
What is already known about this topic? Adults with disabilities experience higher levels of mental health conditions and substance use than do adults without disabilities.
What is added by this report?
During February-March 2021, 64.1% of surveyed U.S. adults with disabilities reported adverse mental health symptoms or substance use; past-month substance use was higher than that among adults without disabilities (40.6% versus 24.5%, respectively). Among adults with a diagnosis of mental health or substance use conditions, adults with disabilities more frequently (43% versus 35%) reported pandemic-related difficulty accessing related care and medications.
What are the implications for public health practice?
During public health emergencies, including the COVID-19 pandemic, enhanced mental health and substance use screening among adults with disabilities and improved access to related health care services are critical. Breakthrough infections were defined as new cases among persons who were fully vaccinated on the day of specimen collection. Hospitalizations among persons with breakthrough infection were defined as new hospital admissions among persons fully vaccinated on the reporting day. The total adult state population that was fully vaccinated and unvaccinated ¶ was assessed for each day and stratified by age group (18-49 years, 50-64 years, and ≥65 years). Persons who were partially vaccinated were excluded from analyses. For each week and age group, the rates of new cases and hospitalizations were calculated among fully vaccinated and unvaccinated persons, by respectively dividing the counts for each group by the fully vaccinated and unvaccinated persondays in that week. Age-adjusted VE each week was estimated † Final dose was the second dose for Pfizer-BioNTech and Moderna vaccines, first dose for Janssen vaccine. § https://www.governor.ny.gov/news/governor-cuomo-announces-new-yorkers-30-years-age-and-older-will-be-eligible-receive-covid-19 ¶ The total adult state population that was unvaccinated was calculated as the total U.S. Census population, minus fully or partially vaccinated persons. Persons who were partially vaccinated were defined as those who initiated a vaccine series but did not complete it or were within 14 days after completion. This report has been corrected and republished. Please click here to view the corrected report and click here to view the detailed changes to the report.
as the population-weighted mean of the age-stratified VE.** The interval between completing vaccination and positive SARS-CoV-2 test result date was summarized using the median, interquartile range (IQR), and percentage tested ≥7 days from being fully vaccinated. † † The ratio of hospitalizations to cases was computed for each vaccination group to understand the relative severity of cases. Statistical testing was not performed because the study included the whole population of interest and was not a sample.
By  (Table). Most (98.1%) new cases among fully vaccinated persons occurred ≥7 days after being classified fully vaccinated (median = 85 days; IQR = 58-113). During May 3-July 25, case rates among fully vaccinated persons were generally similar across age groups, as ** For both outcomes, VE at each week and age group was calculated as 1-(Rate vaccinated / Rate unvaccinated ). † † The percentage tested ≥7 days from being fully vaccinated was included to inform possible undiagnosed infection before full vaccination was achieved.
were case rates among unvaccinated persons, declining through the end of June before increasing in July (Figure 1). Weekly estimated VE against new laboratory-confirmed infection during May 3-July 25 for all age groups generally declined, ranging from 90.6% to 74.6% for persons aged 18-49 years, 93.5% to 83.4% for persons aged 50-64 years, and 92.3% to 88.9% for persons aged ≥65 years. During May 3-July 25, the overall, age-adjusted VE against infection declined from 91.7% to 79.8% (Figure 1) (Table). A total of 1,271 new COVID-19 hospitalizations (0.17 per 100,000 person-days) occurred among fully vaccinated adults, compared with 7,308 (2.03 per 100,000 person-days) among unvaccinated adults (Table). Hospitalization rates generally declined through the week of July 5, but increased the weeks of July 12 and July 19, and were higher among fully vaccinated and unvaccinated persons aged ≥65 years compared with younger age groups ( Figure 2). Age group-specific estimated VE against hospitalization remained stable, ranging from 90.8% to 97.5% for persons aged 18-49 years, from 92.4% to 97.0% for persons aged 50-64 years, and from 92.3% to 96.1% for persons aged ≥65 years. During May 3-July 25, the overall, age-adjusted VE against hospitalization was generally stable from 91.9% to 95.3% ( Figure 2) (Table). The ratio of hospitalizations to cases was moderately lower among fully vaccinated (13.1 hospitalizations per 100 cases) compared with unvaccinated (19.0 hospitalizations per 100 cases) groups.

TABLE. Vaccination coverage, new COVID-19 cases, and new hospitalizations with laboratory-confirmed COVID-19 among fully vaccinated and unvaccinated adults, and estimated vaccine effectiveness -New York, May 3-July 25, 2021
Week starting § New hospitalizations were determined by a report of a hospital admission with a confirmed COVID-19 diagnosis, entered into the Health Electronic Response Data System, which includes a statewide, daily electronic survey of all inpatient facilities in New York. ¶ Persons were determined to be fully vaccinated following 14 days after final vaccine-series dose receipt, per the Citywide Immunization Registry and the New York State Immunization Information System, which collect and store all COVID-19 vaccine receipt data by providers for persons residing in New York City and the rest of New York, respectively.

Discussion
In this study, current COVID-19 vaccines were highly effective against hospitalization (VE >90%) for fully vaccinated New York residents, even during a period during which prevalence of the Delta variant increased from <2% to >80% in the U.S. region that includes New York, societal public health restrictions eased, § § and adult full-vaccine coverage in § § https://covid.cdc.gov/covid-data-tracker/#variant-proportions

Summary
What is already known about this topic?

Real-world studies of population-level vaccine effectiveness against laboratory-confirmed SARS-CoV-2 infection and COVID-19 hospitalizations are limited in the United States.
What is added by this report?
During May 3-July 25, 2021, the overall age-adjusted vaccine effectiveness against hospitalization in New York was relatively stable (91.9%-95.3%). The overall age-adjusted vaccine effectiveness against infection for all New York adults declined from 91.7% to 79.8%.
What are the implications for public health practice?
These findings support the implementation of multicomponent approach to controlling the pandemic, centered on vaccination, as well as other prevention strategies such as masking and physical distancing.
The findings from this study are consistent with those observed in other countries. Israel has reported 90% VE for the Pfizer-BioNTech vaccine against hospitalization; however, a decline in VE against new diagnosed infections occurred during June 20-July 17 (decreasing to <65%) (5). Another study in the United Kingdom found higher VE against infection with the Delta variant for Pfizer-BioNTech (88%), which was lower than VE against the B.1.1.7 (Alpha) variant (94%) (6).
The factors driving the apparent changes in VE, including variations by age, are uncertain. Changes in immune protection from current vaccine product dosing regimens are under investigation, ¶ ¶ with additional doses being considered (7). Increased Delta variant viral load might underpin its increased transmissibility and could potentially lead to reduced vaccineinduced protection from infection (8). Further, variations from clinical trial findings could be because the trials were conducted during a period before the emergence of new variants and when nonpharmaceutical intervention strategies (e.g., wearing masks and physically distancing) were more stringently implemented, potentially lessening the amount of virus to which persons were exposed. Other factors that could influence VE include indirect protective effects of unvaccinated persons by vaccinated persons and an increasing proportion of unvaccinated persons acquiring some level of immunity through infection (9).
The findings in this report are subject to at least six limitations. First, although limiting the analysis period to after universal adult vaccine eligibility and age stratification likely helped to reduce biases, residual differences between fully vaccinated and unvaccinated groups have the potential to reduce estimated VE. Second, the analysis excluded partially vaccinated persons, to robustly assess VE for fully vaccinated ¶ ¶ https://www.medrxiv.org/content/ 10.1101/2021.07.28.21261159v1 compared with that of unvaccinated persons. A supplementary sensitivity analysis that included partially vaccinated persons as unvaccinated yielded conservative VE for laboratory-confirmed infection (declining from 88.7% to 72.1%) and for hospitalizations (ranging from 89.7% to 93.0%). Third, exact algorithms were used to link databases; some persons were possibly not linked because matching variables were entered differently in the respective systems. Fourth, this study did not estimate VE by vaccine product, and persons were categorized fully vaccinated at 14 days after final dose, per CDC definitions; however, the Janssen vaccine might have higher efficacy at 28 days.*** Given that Janssen vaccine recipients accounted for 9% of fully vaccinated persons and the observed time period from full vaccination to infection (median 85 days), this would minimally affect the findings. Fifth, information on reasons for testing and hospitalization, including symptoms, was limited. However, a supplementary analysis found that among 1,271 fully vaccinated adults and 7,308 unvaccinated adults, 545 (42.9%) and 4,245 (58.1%), respectively, were reported to have been admitted for COVID-19 by hospital staff members using nonstandardized definitions. A sensitivity analysis of hospitalization VE limited to those admitted for COVID-19, found similar results (VE range = 93.9%-97.4%), suggesting that the extent of bias was limited. Finally, data were too sparse to reliably estimate VE for COVID-19-related deaths.
This study's findings suggest currently available vaccines have high effectiveness for preventing laboratory-confirmed SARS-CoV-2 infection and COVID-19 hospitalization. However, VE against infection appears to have declined in recent months in New York, coinciding with a period of easing societal public health restrictions † † † and increasing Delta variant circulation (8). These findings support a multipronged approach to reducing new COVID-19 hospitalizations and cases, centered on vaccination, and including other approaches such as masking and physical distancing.
Evaluations of authorized mRNA COVID-19 vaccines (Pfizer-BioNTech and Moderna) have consistently demonstrated high VE across diverse populations (1,5). Because COVID-19 vaccines were initially authorized in the United States in December 2020, evaluations of real-world effectiveness have been subject to a short period of postvaccination follow-up. Monitoring durability of protection after COVID-19 vaccination can help determine whether booster vaccines might be indicated, particularly with continued emergence of new variants that might overcome vaccine-induced immunity. In real-world settings, durability of protection has commonly been measured by comparing the odds of vaccination in laboratory-confirmed case-patients and control-patients who tested negative for infection, by time since vaccination (6,7).
During March 11-July 14, 2021, adults aged ≥18 years admitted to 21 hospitals in 18 states were included in an analysis of durability of vaccine-induced protection. Casepatients had COVID-19-like illness † and had received a positive SARS-CoV-2 reverse transcription-polymerase chain reaction (RT-PCR) or antigen test result. A first group of hospital-based control-patients had COVID-19-like illness and had negative SARS-COV-2 results by all tests, including at least one RT-PCR test. A second hospital-based control group of patients without COVID-19-like illness (and therefore unlikely to be hospitalized for COVID-19-like illness) was also enrolled (4). This second control group also received negative SARS-CoV-2 results by all tests, including at least one RT-PCR test. Eligibility for enrollment as a case-patient or one of these controls required SARS-CoV-2 testing within 10 days of symptom onset and hospital admission within 14 days of symptom onset. Final case/control status was determined using clinical testing results and central laboratory RT-PCR testing of upper respiratory specimens (nasal swabs or saliva) performed at a central laboratory (Vanderbilt University Medical Center, Nashville, Tennessee) (4). Specimens positive for SARS-CoV-2 with cycle threshold values <32 were sent to University of Michigan (Ann Arbor, Michigan) for whole genome sequencing and SARS-CoV-2 lineage determination (4).
Patients or their proxies were interviewed about baseline demographic characteristics, clinical history (including COVID-19-like signs or symptoms experienced and date of illness onset), and history of COVID-19 vaccination. Vaccine was considered to have been administered if vaccination dates and product names were verified through medical records, state immunization registries, vaccination record cards, or provider or pharmacy records, or if plausibly reported by patient or proxy with date and location of vaccination. A patient was considered to be fully vaccinated if both doses of an authorized mRNA COVID-19 vaccine were administered, with the second dose received ≥14 days before illness onset. § Participants were excluded from this analysis if they received only 1 dose of an mRNA COVID-19 vaccine, received 2 doses with the second dose <14 days before illness onset, received a non-mRNA COVID-19 vaccine, or received mixed products of an mRNA COVID-19 vaccine (i.e., a different product for each dose).
Vaccine effectiveness against COVID-19-associated hospitalization was estimated using logistic regression, comparing the odds of being fully vaccinated versus unvaccinated between case-patients and controls (including both control groups) using the equation VE = 100 × (1 -odds ratio) (1). VE over the full surveillance period was assessed, as well as among those with illness onset during March-May and June-July 2021, because of increased circulation of Delta variants in the United States during the latter period (8). Models were adjusted for potential confounders, including admission date (biweekly intervals), U.S. Department of Health and Human Services region, age, sex, and race/ethnicity. Time-varying VE models were then constructed. First, a binary model was constructed by adding a categorical term (2-12 weeks versus 13-24 weeks) for interval from receipt of the second vaccine § The date of illness onset was used for cases and controls with COVID-19-like illness with median value imputed if missing. For controls without COVID-19like illness, the date of admission minus the median number of days between illness onset and admission for patients with COVID-19 was used for a date of illness onset, also referred to as "illness onset" for this report.
dose (among vaccinated participants) and illness onset. Unvaccinated patients were assigned values of zero days since vaccination. In additional analyses, other specifications of time were considered, including using linear and natural cubic spline terms. Bootstrapping with 1,000 replications was used to estimate 95% CIs. Subgroup analyses included adults aged ≥65 years, patients with immunocompromising conditions, ¶ and patients with three or more categories of chronic medical conditions. A sensitivity analysis was also performed including each of the two control groups in models rather than combining them. Significance of association between VE and time since vaccination was assessed using a likelihood-ratio chi-square test with p-values <0.05 considered statistically significant. Analyses were conducted using R statistical software (version 4.0.3; R Foundation). This activity was determined to be public health surveillance by each participating site and CDC and was conducted consistent with applicable federal law and CDC policy.** After excluding 722 ineligible patients (461 who were not fully vaccinated or unvaccinated, 127 who received a non-mRNA COVID-19 vaccine or mixed vaccines, and 134 who did not meet other inclusion criteria), 3,089 patients were included in the final analysis (1,194 case-patients and 1,895 in the combined control groups) (Table). The median patient age was 59 years (interquartile range = 46-69 years), 48.7% were female, 56.7% were non-Hispanic White, and 21.1% had an immunocompromising condition. Among case-patients, 141 (11.8%) were fully vaccinated as were 988 (52.1%) controls. Among 454 case-patient specimens with SARS-CoV-2 lineage determined, 242 (53.3%) were identified as Alpha and 74 (16.3%) as Delta (Figure 1). Delta variants became the dominant virus in mid-June. Overall VE against hospitalization for COVID-19 was 86% (95% CI = 82%-88%) over the full surveillance period, including 90% (95% CI = 87%-92%) among patients without immunocompromising conditions and 63% (95% CI = 44%-76%) among patients with immunocompromising conditions. VE among patients with illness onset during March-May was 87% (95% CI = 83%-90%), and among those with illness onset during June-July was 84% (95% CI = 79%-89%). In models considering time since vaccination, VE was 86% (95% CI = 82%-90%) ¶ Immunocompromising conditions included having one or more of the following: active solid organ cancer ( (Figure 2). In sensitivity analyses, results were similar using individual control groups and combined controls.

Discussion
In a multistate network that enrolled adults hospitalized during March-July 2021, effectiveness of 2 doses of mRNA vaccine against COVID-19-associated hospitalization was sustained over a follow-up period of 24 weeks (approximately 6 months). These findings of sustained VE were consistent among subgroups at highest risk for severe outcomes from COVID-19, including older adults, adults with three or more chronic medical conditions, and those with immunocompromising conditions. Overall VE in adults with immunocompromising conditions was lower than that in those without immunocompromising conditions but was sustained over time in both populations.
These data provide evidence for sustained high protection from severe COVID-19 requiring hospitalization for up to 24 weeks among fully vaccinated adults, which is consistent with data demonstrating mRNA COVID-19 vaccines have the capacity to induce durable immunity, particularly in limiting the severity of disease (9,10). Alpha variants were the predominant viruses sequenced, although Delta variants became dominant starting in mid-June, consistent with national surveillance data (8). Because of limited sequenced virus, Delta-specific VE was not assessed. VE was similar during June-July when circulation of Delta increased in the United States compared with VE during March-May when Alpha variants predominated, although further surveillance is needed.
The findings in this report are subject to at least six limitations. First, the follow-up period was limited to approximately 24 weeks since receipt of full vaccination because of the recent authorization of mRNA COVID-19 vaccines in the United States. Additional analyses with longer duration of follow-up since vaccination are warranted. Second, effectiveness over time from authorized non-mRNA COVID-19 vaccines, including Janssen's (Johnson & Johnson) vaccine product, was not assessed because of limited use of this vaccine during the surveillance period. Third, time-varying VE was not assessed

Summary
What is already known about this topic?
COVID-19 mRNA vaccines provide strong protection against severe COVID-19; however, the duration of protection is uncertain.
What is added by this report?
Among 1,129 patients who received 2 doses of a mRNA vaccine, no decline in vaccine effectiveness against COVID-19 hospitalization was observed over 24 weeks. Vaccine effectiveness was 86% 2-12 weeks after vaccination and 84% at 13-24 weeks. Vaccine effectiveness was sustained among groups at risk for severe COVID-19.
What are the implications for public health practice?
mRNA vaccine effectiveness against COVID-19-associated hospitalizations was sustained over 24 weeks; ongoing monitoring is needed as new SARS-CoV-2 variants emerge. To reduce hospitalization, all eligible persons should be offered COVID-19 vaccination.
by lineage because of sample size. Fourth, residual confounding might have been present, although the analysis adjusted for potential confounders, including calendar time and patient age. Fifth, this analysis did not consider VE over time among persons aged <18 years or partially vaccinated persons. Finally, the current analysis only included hospitalized adults and did not include persons with asymptomatic SARS-CoV-2 infection or COVID-19 who did not require hospitalization. Protection against severe COVID-19 resulting in hospitalization was sustained through 24 weeks after vaccination with mRNA COVID-19 vaccines. To reduce their risk for hospitalization, all eligible persons should be offered COVID-19 vaccination. Continued monitoring of VE against infection and severe disease is needed as the elapsed time since vaccination increases and new SARS-CoV-2 variants emerge.
All authors have completed and submitted the International Committee of Medical Journal Editors form for disclosure of potential conflicts of interest. Samuel M. Brown reports personal fees from Hamilton, institutional fees from Faron Pharmaceuticals and Sedana, grants from Janssen, the National Institutes of Health (NIH), and the Department of Defense (DoD), book royalties from Oxford University and Brigham Young University, outside the submitted work. Jonathan D. Casey reports grants from NIH, outside the submitted work. Steven Y. Chang was a speaker for La Jolla Pharmaceuticals in 2018 and consulted for PureTech Health in 2020. James D. Chappell reports grants from NIH during the conduct of the study. Matthew C. Exline reports support from Abbott Labs for sponsored talks, outside the submitted work. D. Clark Files reports personal consultant fees from Cytovale and is a data and safety monitoring board (DSMB) member from Medpace, outside the submitted work. Adit A. Ginde reports grants from NIH, DoD, AbbVie, and Faron Pharmaceuticals, outside the submitted work. Michelle N. Gong reports grants from NIH and the Agency for Healthcare Research and Quality (AHRQ), DSMB membership fees from Regeneron, and personal fees from Philips Healthcare, outside the submitted work. Carlos G. Grijalva reports consultancy fees from Pfizer, Merck, and Sanofi-Pasteur; grants from Campbell Alliance/Syneos Health, NIH, the Food and Drug Administration, AHQR, and Sanofi, outside the submitted work. David N. Hager reports salary support from Incyte Corporation, the Marcus Foundation, and EMPACT Precision Medicine via Vanderbilt University Medical Center, outside the submitted work. Natasha Halasa reports grants and nonfinancial support from Sanofi, and Quidel outside the submitted work. Daniel J. Henning reports personal consultant fees from Cytovale and Opticyte. Akram Khan reports grants from United Therapeutics, Johnson & Johnson, 4D Medical, Lung LLC, and Reata Pharmaceuticals, outside the submitted work. Adam S. Lauring reports personal fees from Sanofi and Roche, outside the submitted work. Christopher J. Lindsell reports grants from NIH, DoD, and the Marcus Foundation; contract fees from bioMerieux, Endpoint LLC, and Entegrion Inc, outside the submitted work and has a patent for risk stratification in sepsis and septic shock issued. Emily T. Martin reports personal fees from Pfizer and grants from Merck, outside the submitted work. Arnold S. Monto reports consulting fees from Sanofi-Pasteur and Seqirus outside the submitted work. Ithan D. Peltan reports grants from NIH and Janssen Pharmaceuticals and institutional support from Asahi Kasei Pharma and Regeneron, outside the submitted work. Todd W. Rice reports personal fees from Cumberland Pharmaceuticals, Inc. and personal fees from Avisa Pharma, LLC and Sanofi, outside the submitted work. Wesley H. Self reports consulting fees from Aeprio Pharmaceuticals and Merck outside the submitted work. No other potential conflicts of interest were disclosed.

Effectiveness of Pfizer-BioNTech and Moderna Vaccines in Preventing SARS-CoV-2 Infection Among Nursing Home Residents Before and During Widespread Circulation of the SARS-CoV-2 B.1.617.2 (Delta) Variant -National Healthcare Safety Network, March 1-August 1, 2021
Srinivas Nanduri, MD 1, *; Tamara  Nursing home and long-term care facility residents live in congregate settings and are often elderly and frail, putting them at high risk for infection with SARS-CoV-2, the virus that causes COVID-19, and severe COVID-19-associated outcomes; therefore, this population was prioritized for early vaccination in the United States (1). Following rapid distribution and administration of the mRNA COVID-19 vaccines (Pfizer-BioNTech and Moderna) under an Emergency Use Authorization by the Food and Drug Administration (2), observational studies among nursing home residents demonstrated vaccine effectiveness (VE) ranging from 53% to 92% against SARS-CoV-2 infection (3)(4)(5)(6). However, concerns about the potential for waning vaccine-induced immunity and the recent emergence of the highly transmissible SARS-CoV-2 B.1.617.2 (Delta) variant † highlight the need to continue to monitor VE (7). Weekly data reported by the Centers for Medicaid & Medicare (CMS)-certified skilled nursing facilities or nursing homes to CDC's National Healthcare Safety Network (NHSN) § were analyzed to evaluate effectiveness of full vaccination (2 doses received ≥14 days earlier) with any of the two currently authorized mRNA COVID-19 vaccines during the period soon after vaccine introduction and before the Delta variant was circulating (pre-Delta [March 1-May 9, 2021]), and when the Delta variant predominated ¶ (Delta [June 21- August 1, 2021]). Using 17,407 weekly reports from 3,862 facilities from the pre-Delta period, adjusted effectiveness against infection for any mRNA vaccine was 74.7% (95% confidence interval [CI] = 70.0%-78.8%). Analysis using 33,160 weekly reports from 11,581 facilities during an intermediate period  found that the adjusted effectiveness was 67.5% (95% CI = 60.1%-73.5%). Analysis using 85,593 weekly reports from 14,917 facilities during the Delta period found that the adjusted effectiveness * These authors contributed equally to this report. † https://www.cdc.gov/coronavirus/2019-ncov/variants/delta-variant.html § https://www.cdc.gov/nhsn/ltc/covid19/index.html ¶ https://covid.cdc.gov/covid-data-tracker/#variant-proportions was 53.1% (95% CI = 49.1%-56.7%). Effectiveness estimates were similar for Pfizer-BioNTech and Moderna vaccines. These findings indicate that mRNA vaccines provide protection against SARS-CoV-2 infection among nursing home residents; however, VE was lower after the Delta variant became the predominant circulating strain in the United States. This analysis assessed VE against any infection, without being able to distinguish between asymptomatic and symptomatic presentations. Additional evaluations are needed to understand protection against severe disease in nursing home residents over time. Because nursing home residents might remain at some risk for SARS-CoV-2 infection despite vaccination, multiple COVID-19 prevention strategies, including infection control, testing, and vaccination of nursing home staff members, residents, and visitors, are critical. An additional dose of COVID-19 vaccine might be considered for nursing home and long-term care facility residents to optimize a protective immune response.
Effectiveness of mRNA COVID-19 vaccines against laboratory-confirmed SARS-CoV-2 infection among nursing home residents was evaluated using data reported to NHSN. CMScertified nursing homes are required to report aggregate weekly numbers of new laboratory-confirmed SARS-CoV-2 infections among residents, by vaccination status (product and number of doses received), to NHSN. Vaccination status of cases was categorized as 1) unvaccinated (no COVID-19 vaccine doses); 2) fully vaccinated with an mRNA vaccine (2 doses ≥14 days before collection of a SARS-CoV-2-positive specimen), and 3) "other" (single dose of mRNA or Janssen [Johnson & Johnson] vaccine or received unspecified vaccines). Nursing homes also reported weekly on the number of residents by vaccination status; reporting on resident vaccination status was voluntary during the pre-Delta period but was required by CMS starting on June 6, 2021. Facility-level weekly records for the analysis combined case counts by vaccination status in each week with the weekly number of residents by vaccination status 2 weeks previously. This ensured that residents were counted as fully vaccinated only after ≥14 days from receipt of a second dose. Weekly reports of case counts were excluded if a facility did not report resident counts by vaccination status for the corresponding week 2 weeks earlier.
Records from facilities with case data during March 1-August 1, 2021, and the corresponding data on resident vaccination status during February 15-July 18, 2021, were combined for an overall 22-week study period. During the study period, 15,254 facilities sent 330,864 weekly reports with case counts to NHSN; of these, 15,236 facilities (99.9%) sent 144,334 (43.6%) weekly reports with counts of residents by vaccination status.
A generalized linear mixed effects model was used with a zero-inflated Poisson distribution (used to model data that have an excess of zero counts) for case counts by vaccination status, offset by resident counts, to estimate the ratio of infection rates among fully vaccinated and unvaccinated residents. To account for variability across sites, facility was included as a random effect. Because of potential for confounding by time, calendar week was modeled as a fixed effect covariate. Nonlinearity of infection rates over calendar weeks was modeled with cubic splines. To evaluate the effect of circulating SARS-CoV-2 variants on VE, the study period was stratified into three periods: 1) pre-Delta (March 1-May 9); 2) intermediate, the period when Delta circulation was documented but not predominant (May 10-June 20); and 3) Delta, when ≥50% of SARS-CoV-2 viruses sequenced were the Delta variant (June 21-August 1), with an interaction term between this categorical time variable and vaccination status to obtain VE estimates for each period. The following characteristics were evaluated as potential confounders of VE: 1) facility-level cumulative SARS-CoV-2 infection rates combined for staff members and residents from May 8, 2020, through the week of reporting; 2) weekly local county incidence of SARS-CoV-2 infections; and 3) CDC Social Vulnerability Index score** for each facility's county. The change-in-estimate criterion for the regression coefficient with a 10% cutoff was used to evaluate covariates; none met this criterion. VE was estimated as 1 minus the rate ratio multiplied by 100, adjusted for calendar week and facility as a random effect. VE for the "other" category is not presented because this group combines different categories, and estimates would not be meaningful. Data analysis was conducted using SAS (version 9.4; SAS Institute) and R (version 4.0.4; R Foundation); statistical significance was defined as p<0.05. This activity was reviewed by CDC and was conducted consistent with federal laws and institutional policies. † † After applying exclusion criteria and combining facility-level weekly case and corresponding resident counts, the analysis included 136,160 reports from 14,997 facilities (median of eight reports per facility; interquartile range = 6-10), with 3,862 (25.8%) (Table). During the Delta period, adjusted VE against infection among those fully vaccinated was 53.1% for any mRNA vaccine, 52.4% for Pfizer-BioNTech, and 50.6% for Moderna. VE estimates for the Delta period were significantly lower than those for the pre-Delta period (p<0.001). VE point estimates during the intermediate period were lower than those during the pre-Delta period; however, the estimates were not significantly different (p = 0.06) ( Table).

Discussion
Analysis of nursing home COVID-19 data from NHSN indicated a significant decline in effectiveness of full mRNA COVID-19 vaccination against laboratory-confirmed SARS-CoV-2 infection, from 74.7% during the pre-Delta period (March 1-May 9, 2021) to 53.1% during the period when the Delta variant predominated in the United States. This study could not differentiate the independent impact of the Delta variant from other factors, such as potential waning of vaccine-induced immunity. Further research on the possible impact of both factors on VE among nursing home residents is warranted. Because nursing home residents might remain at some risk for SARS-CoV-2 infection despite vaccination, multipronged COVID-19 prevention strategies, including infection control, § § testing, and vaccination of nursing home staff members, residents, and visitors are critical.
These results (pre-Delta 74.7%; Delta 53.1%) fall within the range of findings from other studies of COVID-19 mRNA VE in nursing home residents conducted before the Delta variant was prevalent, with estimates against infection ranging from 53% to 92% (3)(4)(5)(6). Variability in VE estimates across studies can result from differences in underlying populations, SARS-CoV-2 testing practices and diagnostics, prevalence of previous infections, analytic methods, and virus variant strains in circulation.
Nursing home residents, who are often elderly and frail, might have a less robust response to vaccines, and are at higher risk for infection with SARS-CoV-2 and for severe . In addition, nursing home residents were among the earliest groups vaccinated in the United States; thus, if vaccineinduced immunity does wane over time, this decrease in VE might first be observed among nursing home residents. Because increased U.S. circulation of the Delta variant coincided with a period ≥6 months after vaccine introduction, the extent to which reduced vaccine protection against Delta and potential waning immunity contributed to the lower VE in the Delta period could not be determined by this study.
Nursing homes were aggressive in case ascertainment because of guidelines recommending weekly point prevalence surveys if a single SARS-CoV-2 infection in a staff member or resident was identified. ¶ ¶ This analysis assessed VE against any infection, without being able to distinguish between asymptomatic and symptomatic infections. Additional evaluations are needed to understand protection against severe disease in nursing home residents over time.
The findings in this report are subject to at least five limitations. First, resident-level demographic or clinical data were not ¶ ¶ https://www.cms.gov/files/document/qso-20-38-nh.pdf reported to NHSN. Therefore, the analysis could not control for potential confounders, such as age, presence of underlying health conditions, or the influence of previous SARS-CoV-2 infections on VE. Second, vaccination dates were not available and time since vaccination could not be measured to evaluate potential waning of protection. Third, staff member vaccination data were not sufficiently complete to assess as a potential confounder. Fourth, before June 7, 2021, weekly reporting of resident vaccination status was voluntary, and missing data limited inclusion of facility records during this period. Although the magnitude of potential bias introduced by missing data could not be assessed, a bias indicator analysis was conducted, which indicated that VE was likely underestimated during the pre-Delta period (COVID-19 Vaccine Effectiveness Team, CDC, unpublished data, 2021). Finally, the study assessed only nursing home residents and is not generalizable to the broader population.
Both Pfizer-BioNTech and Moderna mRNA vaccines were highly effective in preventing SARS-CoV-2 infection in nursing home residents early after vaccine introduction. However, the effectiveness among this population in recent months has been significantly lower. To prevent transmission of SARS-CoV-2 in nursing homes, these findings highlight the critical importance of COVID-19 vaccination of staff members, residents,

Summary
What is already known about this topic?
Early observational studies among nursing home residents showed mRNA vaccines to be 53% to 92% effective against SARS-CoV-2 infection.
What is added by this report?
Two doses of mRNA vaccines were 74.7% effective against infection among nursing home residents early in the vaccination program (March-May 2021). During June-July 2021, when B.1.617.2 (Delta) variant circulation predominated, effectiveness declined significantly to 53.1%.
What are the implications for public health practice?
Multicomponent COVID-19 prevention strategies, including vaccination of nursing home staff members, residents, and visitors, are critical. An additional dose of COVID-19 vaccine might be considered for nursing home and long-term care facility residents to optimize a protective immune response. and visitors and adherence to rigorous COVID-19 prevention strategies. An additional dose of COVID-19 vaccine might be considered for nursing home and long-term care facility residents to optimize a protective immune response.*** Corresponding author: Srinivas Nanduri, snanduri@cdc.gov.
On August 24, 2021, this report was posted as an MMWR Early Release on the MMWR website (https://www.cdc.gov/mmwr).
During December 14, 2020-April 10, 2021, data from the HEROES-RECOVER Cohorts,* a network of prospective cohorts among frontline workers, showed that the Pfizer-BioNTech and Moderna mRNA COVID-19 vaccines were approximately 90% effective in preventing symptomatic and asymptomatic infection with SARS-CoV-2, the virus that causes COVID-19, in real-world conditions (1,2). This report updates vaccine effectiveness (VE) estimates including all COVID-19 vaccines available through August 14, 2021, and examines whether VE differs for adults with increasing time since completion of all recommended vaccine doses. VE before and during SARS-CoV-2 B.1.617.2 (Delta) variant predominance, which coincided with an increase in reported COVID-19 vaccine breakthrough infections, were compared (3,4).
Methods for the HEROES-RECOVER Cohorts have been published previously (1,2,5). Health care personnel, first responders, and other essential and frontline workers in eight U.S. locations across six states were tested weekly for SARS-CoV-2 infection by reverse transcription-polymerase chain reaction (RT-PCR) † and upon the onset of any COVID-19-like illness. Weeks when the Delta variant accounted for ≥50% of viruses sequenced, based on data from each respective location, were defined as weeks of Delta variant predominance. Vaccination was documented by self-report and verified by provision of vaccine cards or extraction from electronic medical records or state immunization registries. Among 4,217 participants,3,483 (83%) were vaccinated; 2,278 (65%) received Pfizer-BioNTech, 1,138 (33%) Moderna, and 67 (2%) Janssen (Johnson & Johnson) COVID-19 vaccines. Cox proportional hazards models were used to calculate ratios of unvaccinated to fully vaccinated (≥14 days after receipt of all recommended COVID-19 vaccine doses) infection rates, Differences in the percentages of infections by vaccination status were calculated using chi-square tests for categorical variables and Kruskal-Wallis tests for medians; p-values <0.05 were considered statistically significant. Age-adjusted rolling 7-day SARS-CoV-2 infection and hospitalization rates were estimated by vaccination status.** Using convenience samples, WGS lineage data from all available sequencing results (6,752) † † and Ct values from diagnostic qualitative RT-PCR assays targeting two genes (SARS-CoV-2 nucleocapsid [SC2N; 5,179], ORF1ab [1,041], and N [1,062]) from two laboratories were reported over time by vaccination status. Analyses were conducted using SAS (version 9.4; SAS Institute). This activity was determined by LACDPH's Institutional Review Board (IRB) to be a surveillance activity necessary for public health work and therefore did not require IRB review.
The percentage of fully vaccinated Los Angeles County residents increased from 27% on May 1, 2021, to 51% on July 25, 2021. During the same period, 43,127 cases of SARS-CoV-2 infection among residents aged ≥16 years were reported to LACDPH, including 10,895 (25.3%) in fully vaccinated persons, 1,431 (3.3%) in partially vaccinated persons, and 30,801 (71.4%) in unvaccinated persons (Table). The largest percentages of cases across all groups were among adults aged 30-49 years and 18-29 years, females, and Hispanic persons. Among fully vaccinated persons on July 25, 55.2% had received the Pfizer-BioNTech vaccine, 28.0% had received the Moderna vaccine, and 16.8% had received the Janssen vaccine. Lower percentages of fully vaccinated persons were hospitalized (3.2%), were admitted to an intensive care unit (0.5%), and required mechanical ventilation (0.2%) compared with partially vaccinated persons (6.2%, 1.0%, and 0.3%, respectively) and unvaccinated persons (7.6%, 1.5%, and 0.5%, respectively) (p<0.001). Among hospitalized persons and persons admitted to an intensive care unit, the median age was higher among vaccinated persons ( gov/cancer/uscs/technical_notes/stat_methods/rates.htm). Rolling 7-day incidence was calculated by summing the total number of persons or hospitalizations during a 7-day period and dividing by the total population at the end of the 7-day period. † † WGS lineage data were from all sequencing results reported to LACDPH or sequenced after specimens were referred to LACDPH laboratories.
persons were admitted to a hospital after their SARS-CoV-2 positive test result date compared with unvaccinated persons (4.2%). A lower percentage of deaths (0.2%, 24) occurred among fully vaccinated persons than among partially vaccinated (0.5%, seven) and unvaccinated (0.6%, 176) persons (p<0.001). Death investigations determined that six of the 24 fully vaccinated persons who died had immunocompromising conditions, including HIV infection, cancer (i.e., prostate, pancreatic, lung, or leukemia), and liver transplantation, and that the median age was higher among vaccinated (median = 78 years, IQR = 63.5-87.5 years) and partially vaccinated (median = 74, IQR = 58.0-80.0) persons than among unvaccinated persons (median = 63, IQR = 51.5-79.5) (p = 0.01). Among all Los Angeles County residents, the age-adjusted 7-day incidence and hospitalization rates increased exponentially among unvaccinated, fully vaccinated, and partially vaccinated persons, with the highest rates among unvaccinated persons in late June ( Figure 1). On May 1, in unvaccinated persons, the age-adjusted incidence (35.2 per 100,000 population) was 8.4 times and the age-adjusted hospitalization rate (4.6 per 100,000 population) was 10.0 times the rates in fully vaccinated persons (4.2 and 0.46, respectively). Partially vaccinated persons had a similar incidence (4.1) and hospitalization rate (0.27) as fully vaccinated persons. On July 25, the age-adjusted incidence in unvaccinated persons (315.1) was 4.9 times that in fully vaccinated persons (63.8); the rate among partially vaccinated persons was 46.8. The age-adjusted hospitalization rate in unvaccinated persons (29.4) was 29.2 times the rate in fully vaccinated persons (1.0); the hospitalization rate was similar in partially vaccinated persons (0.90) (Supplementary Table; https://stacks.cdc.gov/view/ cdc/109087).

Discussion
The results of this population-based analysis using linked SARS-CoV-2 infection surveillance and vaccination registry data indicate that fully vaccinated persons aged ≥16 years with SARS-CoV-2 infection were less likely than unvaccinated persons to be hospitalized, to be admitted to an intensive care unit, to require mechanical ventilation, or to die from SARS-CoV-2 infection during a period when the Delta variant became predominant. Although age-adjusted hospitalization rates in partially vaccinated persons were similar to those in fully vaccinated persons, age-adjusted incidences were slightly lower in partially vaccinated persons than in fully vaccinated persons. These data indicate that authorized vaccines protect against SARS-CoV-2 infection and severe COVID-19, even with increased community transmission of the newly predominant Delta variant (2).
The SARS-CoV-2 Delta variant is highly transmissible (3) and became the predominant variant in Los Angeles County during May-July 2021. During this period, SARS-CoV-2 cases and hospitalizations increased substantially, most notably among unvaccinated persons. In May, specimens from fully vaccinated and partially vaccinated persons had higher Ct values for two gene targets compared with unvaccinated persons; however, by July, median Ct values had decreased and were similar in all gene regions in specimens from fully vaccinated, partially vaccinated, and unvaccinated persons. Fully vaccinated * Rolling 7-day incidence was calculated by summing the total number of persons or hospitalizations during a 7-day period and dividing by the total population at the end of the 7-day period. † Persons were considered fully vaccinated ≥14 days after receipt of the second dose in a 2-dose series (Pfizer-BioNTech or Moderna COVID-19 vaccines) or after 1 dose of the single-dose Janssen (Johnson & Johnson) COVID-19 vaccine; partially vaccinated ≥14 days after receipt of the first dose and <14 days after the second dose in a 2-dose series; and unvaccinated <14 days receipt of the first dose of a 2-dose series or 1 dose of the single-dose vaccine or if no vaccination registry data were available.
These findings are similar to those from a recent study showing no difference in Ct values in specimens from vaccinated and unvaccinated persons during a large outbreak (4). Ct values are correlated with the amount of viral nucleic acid present; however, Ct values are an imperfect proxy for viral nucleic acid load, are not standardized across testing platforms, vary § § Additional information on Ct values and their limitations is available: https:// www.idsociety.org/globalassets/idsa/public-health/covid-19/idsa-ampstatement.pdf and https://www.cdc.gov/coronavirus/2019-ncov/lab/faqs.html.

Summary
What is already known about this topic?
Although COVID-19 vaccines are highly effective, some fully vaccinated persons will be infected with SARS-CoV-2.
What is added by this report? by specimen type and time from specimen collection, and should be limited to assessing differences at the population level, not the person level. § § The findings in this report are subject to at least six limitations. First, vaccination data for persons who lived in Los Angeles County at the time of their laboratory-confirmed infection but who were vaccinated outside of California were unavailable, leading to misclassification of their vaccination status; if vaccinated persons without accessible records were considered to be unvaccinated, the incidence in unvaccinated persons could be underestimated. Second, case ascertainment is based on passive surveillance, with known underreporting that might differ by vaccination status. Similarly, screening and testing behaviors might differ among groups. Third, COVID-19-associated hospitalizations were determined based on hospital admission and SARS-CoV-2 test dates alone, leading to the inclusion of incidental hospitalizations that were not associated with COVID-19. Fourth, COVID-19-associated deaths included deaths occurring ≤60 days after a first SARS-CoV-2 positive test date; therefore, some COVID-19associated deaths might have been from other causes (excluding trauma). In addition, certain COVID-19-associated deaths might have been a result of long-term sequelae after 60 days and were not included. Fifth, lineage and Ct values were available only for a convenience sample of SARS-CoV-2 cases. Finally, all the assays used to generate Ct values for comparison were qualitative, and none is approved for use in quantitating the amount of viral nucleic acid present.
The findings in this report are similar to those from recent studies indicating that COVID-19 vaccination protects against severe COVID-19 in areas with increasing prevalence of the SARS-CoV-2 Delta variant (5,6). Efforts to increase COVID-19 vaccination coverage, in coordination with other prevention strategies, are critical to preventing COVID-19-related hospitalizations and deaths. Ongoing surveillance to characterize postvaccination infections, hospitalizations, and deaths will be important to monitor vaccine effectiveness, particularly as new variants emerge.

RT-PCR Ct values
Abbreviations: Ct = cycle threshold; N = nucleocapsid; ORF1ab = open reading frame 1 ab; RT-PCR = reverse transcription-polymerase chain reaction; SC2N = SARS-CoV-2 nucleocapsid. * SARS-CoV-2 infections among Los Angeles County residents aged ≥16 years with whole genome sequencing lineage results (n = 6,752) for fully vaccinated (n = 1,667), partially vaccinated (n = 198), and unvaccinated (n = 4,887) persons. † Whiskers represent minimum and maximum observations; top of box represents the third quartile, bottom represents the first quartile, and box height represents the interquartile range. The midline is the median. § Ct values are correlated with the amount of viral nucleic acid present. Gene targets for RT-PCR testing included the N protein gene region and the ORF1ab polyprotein gene region. The N gene targets were analyzed separately for two laboratories because Ct values are not directly comparable across different testing laboratories; these N gene targets were designated SC2N and N to differentiate between the two participating laboratory partners. Gene targets were selected based on testing platforms used by Los Angeles County Department of Public Health laboratory partners. Analysis of SC2N Ct values is restricted to a Fulgent test result with a Ct value on the same day as person's first positive RT-PCR test result; SC2N gene target values (n = 5,179) are stratified for fully vaccinated (n = 1,248), partially vaccinated (n = 151), and unvaccinated (n = 3,780) persons. Analysis of ORF1ab and N Ct values is restricted to a Valencia Branch Laboratory test result with a Ct value on the same day as person's first positive RT-PCR test result. ORF1ab (n = 1,041) and N (n = 1,062) gene target values are stratified for fully vaccinated (n = 289 and n = 297, respectively), partially vaccinated (n = 36 and n = 41, respectively), and unvaccinated (n = 716 and n = 724, respectively) persons. ¶ Persons were considered fully vaccinated ≥14 days after receipt of the second dose in a 2-dose series (Pfizer-BioNTech or Moderna COVID-19 vaccines) or after 1 dose of the single-dose Janssen (Johnson & Johnson) COVID-19 vaccine; partially vaccinated ≥14 days after receipt of the first dose and <14 days after the second dose in a 2-dose series; and unvaccinated <14 days receipt of the first dose of a 2-dose series or 1 dose of the single-dose vaccine or if no vaccination registry data were available. Cardiopulmonary resuscitation 1 (4.8) to both opioids and illicit benzodiazepines did not respond to naloxone. Although other factors might be involved, such as naloxone dose or administration technique, the opioid effects among these patients might have been reversed; however, these patients possibly experienced additional sedative effects from illicit benzodiazepines. The widespread use of community naloxone programs highlights the importance of calling emergency medical services after administering naloxone, because patients with co-exposure might require additional medical care. The growing use of illicit benzodiazepines requires a better understanding of the synergistic toxicity when these drugs are used along with opioids. Raising awareness among clinical, public safety, and community partners about dangers associated with the use of illicit benzodiazepines, including co-use with opioids, is critical. * Flubromazolam was only detected in two of the cases that also included etizolam. † At least one opioid was identified in 20 cases. More than one opioid might be noted for a given case. The percentages of specific opioids are calculated based on these 20 cases. § The percentages of number of doses of naloxone are calculated based on 16 cases with naloxone administration. Initial naloxone dose was administered either outside of the hospital (by emergency medical services in 10 cases, by bystanders in two cases, and unknown in one case) or in the hospital (three cases). ¶ Indications for initial dose of naloxone were known in 15 of 21 total cases. The percentages of naloxone indication categories are calculated based on these 15 cases. More than one indication might be noted for a given case. ** Response to initial dose of naloxone was known in 13 of 16 naloxone administrations (81.3%). More than one response might be noted for a given case. The percentages of clinical response categories are calculated based on these 13 cases. † † Precipitated withdrawal is medication-induced withdrawal that can cause particularly intense symptoms, including agitation, nausea/vomiting, and muscle aches and pains, among other withdrawal symptoms. Type of event * With 95% confidence intervals indicated with error bars. † Percentages for the specified stressful life events are based on the following questions: 1) "Has child ever been the victim of violence or witnessed violence in their neighborhood?"; 2) "Did child ever live with a parent or guardian who served time in jail or prison after child was born?"; 3) "Did child ever live with anyone mentally ill/depressed?"; 4) Did child ever live with anyone who had a problem with alcohol or drugs?" Having any stressful event was based on having answered "yes" to any of these four questions. The four stressful life event questions are part of a larger battery of questions called adverse childhood experiences. § Poverty status was based on family income and family size, using the U.S. Census Bureau's poverty thresholds. Family income was imputed when missing. ¶ Estimates are based on household interviews of a sample of the civilian, noninstitutionalized U.S. population.
In 2019, 20.7% of children and adolescents in families with incomes <200% of the poverty threshold and 12.6% of children and adolescents in families with incomes ≥200% of the poverty threshold had experienced at least one specified stressful life event. Children and adolescents in families with incomes <200% of the poverty threshold were more likely than children and adolescents in families with incomes ≥200% of the poverty threshold to have been the victim or witnessed violence (8.1% versus 3.5%); lived with someone who had been in jail (8.7% versus 3.5%); lived with a person with problems with mental health or depression (10.1% versus 6.4%); or lived with a person with problems with alcohol or drugs (10.2% versus 6.5%). For more information on this topic, CDC recommends the following link: https://www.cdc.gov/injury/priority/aces.html